Humans are able to digest large quantities of information and are better capable of dealing with ambiguity, something that is a primary feature of the forecasting activity. This approach relies on the expertise of the person or people involved, however. As such, it is typically at least complemented by some form of quantitative forecasting.
Time series forecasting is a method by which forecasts of the future are produced based on past events. There are many different techniques of time series forecasting, such as trend line or method of averages, but each technique is based on extrapolation of past data into the future. This technique is simple, and if future conditions are reasonably similar to past conditions it can also be fairly accurate. The downside of time series forecasting is that it relies on operational stability and a long time series to be truly effective. It does not take into account major shifts in the environment that may otherwise be factored in by a human.
The impact of these variables is then analyzed. Future values for the variables are assumed based on past information, and this in turn leads to the budget figures. Causal methods rely heavily on the user to interpret causal relationships between the input variables and the outputs. This is typically done with a computer, but some degree of human expertise is required. Although causal analysis is the most complex and unwieldy, it is also the most accurate method of forecasting.
Works Cited:
Rabin, Jack. (1992). Handbook of Public Budgeting. Retrieved July 14, 2009 from http://books.google.de/books?id=jyX0oF66J3oC&pg=PA217&lpg=PA217&dq=revenue+forecasting+time+series&source=bl&ots=mSw7SNpEku&sig=GlF6iTgjQ5_0u8RpN6AhFPVWAUE&hl=en&ei=wCJNSv2fKInBsAak1uC1BA&sa=X&oi=book_result&ct=result&resnum=2
